Fructose and glucose can regulate mammalian target of rapamycin complex 1 and lipogenic gene expression via distinct pathways

Although calorically equivalent to glucose, fructose appears to be more lipogenic, promoting dyslipidemia, fatty liver disease, cardiovascular disease, and diabetes. To better understand how fructose induces lipogenesis, we compared the effects of fructose and glucose on mammalian target of rapamycin complex 1 (mTORC1), which appeared to have both positive and negative effects on lipogenic gene expression. We found that fructose acutely and transiently suppressed mTORC1 signaling in vitro and in vivo. The constitutive activation of mTORC1 reduced hepatic lipogenic gene expression and produced hypotriglyceridemia after 1 week of fructose feeding. In contrast, glucose did not suppress mTORC1, and the constitutive activation of mTORC1 failed to suppress plasma triglycerides after 1 week of glucose feeding. Thus, these data reveal fundamental differences in the signaling pathways used by fructose and glucose to regulate lipid metabolism.

Our nation is currently facing an epidemic of metabolic disease in the forms of obesity, diabetes, fatty liver disease, and cardiovascular disease (1)(2)(3). The increase in the prevalence of these disorders is temporally correlated with an increase in the consumption of fructose, prompting appeals for public measures to limit fructose intake (4). However, some of the detrimental effects of fructose are lost after normalization to weight gain, suggesting that the negative effects of fructose are simply due to excess caloric intake rather than any specific property of fructose (5). Thus, the effects of fructose on human health remain a topic of intense debate.
Fructose is nearly identical in structure to glucose, differing solely by the substitution of a hemiketal group for a hemiacetal group. However, fructose appears to be a more effective driver of lipogenesis than glucose as individuals fed fructose, compared with glucose, showed increased de novo lipogenesis (6). This is important because excessive lipogenesis can produce hypertriglyceridemia, fatty liver disease, and potentially even insulin resistance (7,8).
The lipogenic effects of fructose are due in large part to the ability of fructose to induce transcription of the lipogenic enzymes (9,10). Thus, mice fed with a fructose versus glucose diet for 1 week show higher levels of the nuclear transcription factor sterol regulatory element-binding protein 1c (SREBP-1c) 4 (11). SREBP-1c is a master transcriptional regulator, capable of activating all of the enzymes necessary for the synthesis of monounsaturated fatty acids (12). In parallel, mice fed with fructose also show higher levels of the lipogenic transcripts (9,11).
There are several differences in the cellular metabolism of fructose versus glucose (13). Although both glucose and fructose are phosphorylated upon entering the cell, the phosphorylation of glucose is highly sensitive to end-product inhibition, but the phosphorylation of fructose is not. Consequently, the phosphorylation of fructose upon entering the cell is so rapid that ATP levels decrease. The depletion of ATP triggers nucleotide degradation and uric acid formation; consequently, hyperuricemia is observed upon consumption of fructose but not glucose (13)(14)(15). The depletion of ATP by fructose also triggers the activation of adenosine monophosphate-activated protein kinase (AMPK) (16). Interestingly, AMPK inhibits mammalian target of rapamycin complex 1 (mTORC1) (17), which is paradoxical in this context because mTORC1 activates lipogenesis and is required for the induction of SREBP-1c, at least by insulin (18). How fructose interacts with mTORC1 in the control of lipogenesis, however, is not clear.
Here, we studied the effects of glucose and fructose on mTORC1 in vitro and in vivo. We found that fructose and glucose regulate lipid metabolism via distinct signaling pathways. Within minutes, fructose acutely and transiently inhibited mTORC1 signaling in vitro and in vivo; mice unable to suppress mTORC1 showed defects in lipogenic gene expression and hypotriglyceridemia in response to a fructose-enriched diet. In contrast, glucose did not inhibit mTORC1, and mice unable to suppress mTORC1 showed robust lipogenic gene expression and developed steatosis in response to a glucose-enriched diet. These data suggest that mTORC1 signaling interferes with the lipogenic pathways that are activated specifically by fructose feeding.

Results
To determine how fructose and glucose affect signaling, we stimulated primary rat hepatocytes with either fructose or glu-cose for 10 min. As expected from previous studies (16), fructose significantly lowered ATP levels, but glucose did not (Fig.  1A). Consistent with this, p-AMPK, a marker of AMPK activation, was induced in fructose-treated cells (Fig. 1B). In parallel, fructose reduced mTORC1 activation ( Fig. 1C and Fig. S1) as assessed by levels of phosphorylated ribosomal protein S6 kinase (S6K), ribosomal protein S6 (S6), and eukaryotic translation initiation factor 4E-binding protein (4EBP). Fructose did not suppress Akt (Fig. S1), which activates mTORC1 in response to growth factors, such as insulin (19). Furthermore, treatment with the AMPK inhibitor dorsomorphin (20) reduced p-AMPK (21) and attenuated the effects of fructose on S6K phosphorylation. Taken together, these data indicate that fructose inhibits mTORC1, at least in part, via AMPK (Fig. 1D).
In vivo, fructose is partially converted to glucose in the enterocytes, and glucose in turn stimulates the secretion of insulin from the ␤-cells of the pancreas (13). Because insulin induces mTORC1 (22, 23), we asked how fructose feeding affected mTORC1 signaling in vivo. Thus, we gavaged wildtype C57Bl6/J mice with either fructose or glucose and examined Effects of fructose and glucose on mTORC1 signaling hepatic mTORC1 signaling 30 min later. We found that fructose in vivo, as in vitro, lowered p-S6K, p-S6, and p-4EBP (Fig. 2).
To dissect the role of mTORC1 in the acute lipogenic response to fructose feeding, we studied mice with either deletion or constitutive activation of mTORC1 signaling in their livers. To abolish mTORC1 signaling, we utilized mice with liver-specific deletion of raptor, a necessary and specific component of the mTORC1 signaling complex (24): mice homozygous for the floxed allele of the Raptor gene and heterozygous for a transgene encoding the Cre recombinase under the control of the albumin promoter were generated (Albumin-Cre ϩ/Ϫ ; Raptor flox/flox , or L-Raptor KO) and compared with their littermates lacking the Cre transgene (Albumin-Cre Ϫ/Ϫ ; Raptor flox/flox , or Flox).
L-Raptor KO mice and their Flox controls were fasted for 24 h or fasted for 24 h and then refed for 6 h with a fructoseenriched diet (60% fructose by weight). Expression of the lipogenic transcription factors SREBP-1c and carbohydrate response element-binding protein (CHREBP) were measured by real-time PCR. SREBP-1c induces lipogenic gene expression in response to both insulin (25) and carbohydrates (26,27), whereas CHREBP responds primarily to carbohydrates (28,29). In Flox controls, fructose increased expression of Srebp-1c, Chrebp, and the more active transcript of CHREBP, Chrepb-␤ (9,30). It also increased expression of the lipogenic enzymes pyruvate kinase (Lpk), acetyl-CoA carboxylase (Acc), and fatty acid synthase (Fasn) by 1.5-6-fold, although the increase in Acc did not reach significance. Surprisingly, the expression of these genes and their response to fructose were not impaired in L-Raptor KO mice. The only exception was Fasn, which was induced 3-fold by fructose in L-Raptor knockout mice but 6-fold in Flox mice. Stearoyl-CoA desaturase 1 (Scd1) levels did not change in mice of either genotype (Fig. 3A). Thus, lipogenic gene expression was largely preserved in L-Raptor knockout mice after the acute refeeding of fructose.
In parallel, we studied mice with liver-specific activation of mTORC1 due to deletion of tuberous sclerosis 1 (TSC1). These mice were homozygous for the floxed allele of the Tsc1 gene and harbored a transgene encoding the Cre recombinase under the control of the albumin promoter (Albumin-Cre ϩ/Ϫ ; Tsc1 flox/flox , or L-TSC1 KO). The deletion of TSC1 disrupts the inhibitory tuberous sclerosis complex, resulting in the constitutive activation of mTORC1 in the livers of these mice (19). Levels of Srebp-1c, Chrebp, and Chrepb-␤ as well as Acc and Fasn were higher in the livers of L-TSC1 KO mice refed the fructose diet than L-TSC1 KO mice that were fasted (Fig. 3C). However, Srebp-1c, Chrepb-␤, Acc, and Fasn were still 40 -60% lower in the livers of L-TSC1 KO mice refed the fructose diet than Flox mice refed the fructose diet (Fig. 3C). These data indicate that the constitutive activation of mTORC1 partially impairs lipogenic gene expression. At the protein level, nuclear SREBP-1c and CHREBP were normal, but NCoR1 was increased (Fig. 3D). Thus, the activation of mTORC1 increased NCoR1 and reduced expression of Srebp-1c, Chrepb-␤, and Acc, whereas the deletion of mTORC1 signaling did not.
We then studied C57Bl6/J mice either maintained on a chow diet or fed a fructose-enriched or glucose-enriched diet for 1 week. In the glucose-enriched diet, glucose, in the form of corn starch and maltodextrin, was substituted for fructose. As expected from prior studies, glucose and fructose feeding induced lipogenic gene expression at both the mRNA and protein levels (Fig. S2) (11,38).
Interestingly, p-S6 levels were similar in mice fed either the chow, glucose-enriched, or fructose-enriched diet (Fig.  S2B). These data suggest that the suppressive effects of fruc- Eight-to ten-week-old male wildtype C57Bl6/J mice were fasted for 18 h and gavaged with water or 500 mg/kg fructose or glucose. Thirty minutes later, mice were sacrificed, and livers were subjected to immunoblotting. Representative images (left) and the quantified results of two independent cohorts (right) are shown. Bars and error bars correspond to the mean and S.E., respectively. *, p Ͻ 0.05 (Student's t test) versus water treated. T-, total.

Effects of fructose and glucose on mTORC1 signaling
tose on mTORC1 that were observed after 30 min of fructose gavage had been lost after 1 week of fructose feeding. Nonetheless, to define the role of mTORC1 in the control of lipogenic gene expression during prolonged fructose feeding, L-TSC1 KO mice and their Flox controls were fed for 1 week with either a fructose-enriched diet or a glucose-enriched diet.
We found that the activation of mTORC1 reduced lipogenic gene expression to a greater extent on the fructose-enriched diet than on the glucose-enriched diet. On the glucose-enriched diet, the livers of L-TSC1 KO mice showed reduced expression of Fasn but not Srebp-1c, Chrebp-␤, Acc, or Scd1, compared with their Flox controls; indeed, Lpk levels were 2-fold higher. In contrast, on the fructose-enriched diet, the livers of L-TSC1 KO mice showed 20 -70% lower levels of Srebp-1c, Acc, and Fasn than their Flox controls (Fig. 4A). At the protein level, SREBP-1c, CHREBP, and NCoR1, on either diet, were similar in L-TSC1 KO mice and their Flox controls (Fig.  S3). However, ACC, LPK, and SCD1 protein levels were increased in the livers of L-TSC1 KO mice versus their Flox controls on a glucose-enriched diet but not on a fructose-enriched diet, and FASN protein levels were decreased in the livers of L-TSC1 KO mice versus their Flox controls on a fructoseenriched diet but not a glucose-enriched diet (Fig. S3). Furthermore, on the glucose-enriched diet, L-TSC1 KO mice showed normal plasma triglycerides and a 30% increase in hepatic triglycerides compared with their Flox controls (Fig. 4,  B and C). On the fructose-enriched diet, L-TSC1 KO mice showed a 50% decrease in plasma triglycerides but normal hepatic triglycerides (Fig. 4, B and C). Thus, the activation of mTORC1 had negative effects on the lipogenic pathways that were largely restricted to the fructose-enriched diet, reducing the expression of Srebp-1c and Acc mRNA, FASN protein, and plasma triglyceride levels.
To further examine the effects of mTORC1 and glucose or fructose on lipid metabolism, lipidomic analysis was performed in L-TSC1 KO and Flox livers. On both diets, L-TSC1 KO mice showed a higher proportion of C22:4n-6 and C22:5n-6 fatty acids than their Flox controls. However, some changes, such as the decrease in C16:0 and increase in C20:2n-6, were more pro- A and C, hepatic gene expression was measured by real-time PCR. Bars and error bars correspond to the mean and S.E., respectively (n ϭ 4 -7). *, p Ͻ 0.05 (Student's t test) versus fasted mice of the same genotype; #, p Ͻ 0.05 (Student's t test) versus Flox controls sacrificed under the same feeding conditions. B and D, nuclear protein levels were determined by immunoblotting liver nuclear extracts. (N), nuclear.

Effects of fructose and glucose on mTORC1 signaling
nounced on the fructose-enriched diet (Fig. 4D), further illustrating the diet-specific effects of mTORC1 activation.

Discussion
In these studies, we showed that fructose, in contrast to glucose, acutely and transiently inhibits mTORC1. Moreover, mTORC1 impaired the lipogenic response to 1 week of fructose, but not glucose, feeding.
Metabolomic analysis showed that glucose and fructose produce distinct metabolite footprints in primary hepatocytes (Fig.   S4). For example, fructose reduced malonic acid but increased galacturonic acid (Fig. S4). In addition, fructose significantly reduced ATP levels and activated AMPK (Fig. 1, A and B). As the effects of fructose on ATP levels are transient (16), the suppressive effects of fructose on mTORC1 are also transient. Thus, our studies done after 1 week of feeding a fructose-enriched diet showed no defect in mTORC1 activation (Fig. S2B), and other groups, studying longer periods of fructose feeding, have shown increased mTORC1 signaling (39 -43), at least relative to a chow diet.  shown by dotted line). B and C, plasma and hepatic triglycerides were measured using a colorimetric assay. D, hepatic lipidomic analysis was performed using GC-MS; each lipid species is shown as a percentage of the total. Bars and error bars correspond to the mean and S.E., respectively (n ϭ 4 -7). *, p Ͻ 0.05 (Student's t test) versus mice of the same genotype fed the glucose-enriched diet; #, p Ͻ 0.05 (Student's t test) versus Flox mice sacrificed under the same feeding conditions.

Effects of fructose and glucose on mTORC1 signaling
The mTORC1 signaling node has complex effects on lipogenic gene expression. mTORC1 promotes the transcription, maturation, and activity of SREBP-1c (44 -47). Consistent with this, L-Raptor knockout mice showed a decrease in SREBP-1c and its target genes under certain conditions, such as after 1 week of fructose diet feeding (Fig. S5, A and B). At the same time, mTORC1 exerted a negative effect on lipogenic gene expression. In the context of acute refeeding, this negative effect may have been mediated in part via NCoR1. First, nuclear NCoR1 levels were reduced in L-Raptor KO livers and increased in L-TSC1 KO livers (Fig. 3, B and D). Second, hepatic disruption of NCoR1 (32) markedly induced lipogenic gene expression (Fig. S6). These data suggest that fructose may acutely promote lipogenic gene expression by suppressing mTORC1 and NCoR1. Consistent with this, fructose appeared to acutely and transiently suppress NCoR1 (Fig. S6) as well as mTORC1.
It is also likely that the mTORC1 complex inhibits lipogenesis indirectly via Akt (48 -52). Akt activates mTORC1, and mTORC1, via an inhibitory feedback loop, suppresses Akt. However, Akt has additional targets, such as FoxO1 and GSK-3, which have also been implicated in lipogenesis. Thus, mTORC1 may inhibit lipogenesis through these other targets (50,53).
Importantly, the effects of mTORC1 on lipogenesis differ in the context of fructose versus glucose feeding. On the glucoseenriched diet, the constitutive activation of mTORC1 had a largely neutral or even positive effect on lipogenic gene expression and hepatic triglyceride accumulation (Fig. 4, A and C). On the fructose-enriched diet, the constitutive activation of mTORC1 reduced Srebp-1c, Chrebp-␤, and Acc mRNA levels and reduced plasma triglyceride levels (Fig. 4, A and B). Thus, the lipogenic pathways activated by fructose feeding are more sensitive to the inhibitory effects of mTORC1 than those activated by glucose feeding. In summary, our data show that glucose and fructose, despite their caloric equivalence, display profound differences in their interaction with the mTORC1 signaling node, which is an important regulator of lipogenesis as well as proliferation, autophagy, protein, and nucleotide synthesis (18,19).

Animals, diets, and treatments
All procedures were approved by the Institutional Animal Care and Research Advisory Committee at Boston Children's Hospital. Animals were housed in a 12-h light/dark cycle (7 a.m. to 7 p.m.) and received standard chow and water ad libitum unless otherwise indicated. Mice studied were male and sacrificed at 8 -12 weeks of age in the non-fasted state at 2 p.m.
For gavage studies, C57Bl6/J mice (The Jackson Laboratory) were fasted for 18 h. Mice were then gavaged with 200 l of water with or without supplementation with 10% fructose or glucose (500 mg/kg of body weight); 30 min later, mice were sacrificed for tissue collection.
For acute feeding studies, L-TSC1 KO mice and their Flox controls were sacrificed after a 24-h fast or after a 24-h fast followed by refeeding a diet containing 60% fructose (TD00202, Harlan Teklad; 61.8% carbohydrate, 17.7% protein, and 7.2% fat by weight). Alternatively, mice were placed on the fructose-or glucose-enriched diet for 1 week. The glucose-enriched diet was formulated exactly as the fructose-enriched diet except that fructose was substituted with corn starch and maltodextrin (TD120623, Harlan Teklad; 61.5% carbohydrate, 17.7% protein, and 7.2% fat by weight).

Biochemical characterization
Plasma was obtained using EDTA-treated syringes, and colorimetric assays were used for measuring total triglycerides and total cholesterol (Thermo Fisher Scientific). Hepatic lipids were extracted using the Folch method (55,56). Briefly, 100 mg of liver was homogenized in 50 mM NaCl and extracted with chloroform:methanol (2:1). The interphase was washed once with 50 mM NaCl and once with 0.36 M CaCl 2 in 50% methanol. Aliquots of the organic extract were supplemented with Triton X-100 (Sigma) and dried at room temperature. Colorimetric reagents for measuring total cholesterol were added directly to the detergent pellet and read according to the manufacturer's instructions. Lipidomic analysis was carried out as described previously by gas chromatography and mass spectrometry (57).

Primary rat hepatocytes and treatment
All reagents were purchased from Life Technologies unless otherwise indicated. Primary hepatocytes were isolated from 8 -10-week-old male Sprague-Dawley rats (Harlan) and treated as described previously (56,58) with minor modifications. After isolation, cells were suspended in William's E medium containing penicillin-streptomycin, 100 nM glutamine, and 10% fetal bovine serum; 1 ϫ 10 6 cells were placed on rat tail collagen I (BD Biosciences)-coated 6-well plates. Four hours later, cells were washed twice with phosphate-buffered saline (PBS), incubated overnight in low-glucose DMEM (1 g of glucose/liter) containing penicillin-streptomycin (fasting medium) and washed twice more with PBS. Cells were then incubated in fresh low-glucose DMEM with or without supplementation with either 25 mM fructose or 25 mM glucose for 10 min. In some studies, cells were pretreated with 10 M dorsomorphin (Sigma) for 30 min prior to fructose or glucose stimulation.

Gene expression analysis
RNA was isolated using TRIzol, and 1 g was used for reverse transcription (Applied Biosystems) according to the manufacturer's instructions. The resultant cDNA was diluted 10-fold and quantified using Power SYBR Green PCR Master Mix in QuantStudio 6 Flex. Real-time RT-PCR was performed in Effects of fructose and glucose on mTORC1 signaling duplicate or triplicate for each sample. Expression was normalized to the housekeeping gene TATA-binding protein (Tbp). The sequences of primers used for real-time PCR are listed in Table S1.

Immunoblotting
All reagents were purchased from Thermo Fisher Scientific unless otherwise indicated. Whole-cell lysates were prepared by collecting cells in lysis buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Nonidet P-40, 0.5% sodium deoxycholate, 1.0% SDS, 2 mM NaF, 2 mM Na 3 VO 4 , and protease inhibitors (Roche Applied Science)) followed by sonication and centrifugation at 13,000 ϫ g for 10 min (37). Liver lysates were prepared by homogenizing in lysis buffer and then centrifuging at 13,000 ϫ g for 10 min at 4°C. Alternatively, nuclear extracts were prepared by pooling three mice per group and using a commercial kit as described previously (37). Protein was quantified using a BCA assay kit, loaded onto gels for SDS-PAGE, and transferred to a PVDF membrane. After 1 h in SuperBlock blocking buffer, blots were incubated overnight with primary antibodies. Secondary antibody conjugated with horseradish peroxidase and chemiluminescent ECL reagents were used to develop blots. Antibody information is listed in Table S2. ImageJ was used for quantification.

Statistical analysis
For Western blotting, representative images and quantification of two to three independent experiments for both in vitro and vivo experiments are shown. For other experiments, representative results of two to four independent experiments are shown. Differences between groups were assessed by a twotailed unequal variance Student's t test. Bars and error bars correspond to the mean and S.E., respectively.